DeepSeek V4 Flash 0423

DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and...

Deskripsi di bawah ini disediakan oleh OpenRouter dalam bahasa Inggris.

Dirilis

Context window

1M token

Output maksimal

944K token

Input

teks

Output

teks

Penalaran

Ya

Riwayat harga

• input $0.14 → $0.091 -35% · output $0.28 → $0.182 -35%
• input $0.047 → $0.14 +196% · output $0.095 → $0.28 +196%
• input $0.045 → $0.047 +5% · output $0.45 → $0.095 -79%
• input $0.047 → $0.045 -5% · output $0.45 → $0.45 0%
• input $0.05 → $0.047 -5% · output $0.45 → $0.45 0%
• input $0.14 → $0.05 -64% · output $0.28 → $0.45 +61%
• input $0.091 → $0.14 +54% · output $0.182 → $0.28 +54%
• input $0.05 → $0.091 +82% · output $0.45 → $0.182 -60%
• input $0.091 → $0.05 -45% · output $0.182 → $0.45 +147%

Penyedia di OpenRouter

17 endpoint menyediakan model ini. Uptime menunjukkan rasio permintaan berhasil di OpenRouter; harga per 1M token.